Bytesize-SEO

Local-SEO skills for server-rendered, multi-location marketing sites.

Every rule here prevents a failure that is easy to ship and hard to notice — the kind that looks fine in the markup, passes review, and quietly costs you the ranking. The failure is recorded next to the rule on purpose: a rule without its failure gets "simplified" away by the next person who reads it.

Skills

Skill	Covers
seo-playbook	Start here. Cross-cutting rules: head contract, canonicals, host-derived indexing, robots.txt, sitemaps, JSON-LD escaping, thin pages, URLs, verification, deploy traps. Routes to the page skills.
seo-home	Homepage and hero — the page the logo links to. H1 as category term, LCP, above-the-fold conversion, Organization schema.
seo-products	Catalog taxonomy, Product schema, galleries without JavaScript, faceted filters, thin subcategories.
seo-services	Service index and detail pages, staff-editable content, thin-service handling.
seo-locations	LocalBusiness schema, NAP consistency, opening hours, per-city and per-county pages, duplicate-copy avoidance.
seo-about	Founder story, E-E-A-T, trust pages, financing and FAQ.
seo-header	Crawlable nav, CSS-only keyboard-accessible dropdowns, logo markup, sticky-header CLS, skip links.
seo-footer	Footer NAP as a citation surface, hours markup, internal link mesh, decorative logo alt.
seo-accessibility	WCAG 2.2 rules for the same surfaces: keyboard operation, focus visibility under sticky headers, target size, contrast, accessible names, alt-text decisions, form labels, reflow, reduced motion.

Load seo-playbook first, then the skill matching the surface you are touching. The page skills do not repeat the shared rules.

seo-accessibility is the one skill that is not page-specific — load it alongside whichever page skill you are using, any time you touch something interactive.

Install

claude plugin marketplace add nickdemari/Bytesize-SEO
claude plugin install bytesize-seo

Or from a local clone:

claude plugin marketplace add ~/dev/Bytesize-SEO
claude plugin install bytesize-seo

Scope and portability

Snippets are written against Rails 8 / ERB / Propshaft, and examples use a fictional retailer. The snippets are Rails; the rules are not. Canonical-origin discipline, host-derived noindex, thin-page handling, schema @id graphs and NAP consistency apply to any stack that renders HTML on the server.

Two assumptions run through all of it:

1. The site exists to be found. Crawlability, unique per-location copy and structured data are functional requirements, not polish.
2. Content a crawler needs is in the initial HTML response. If a curl of the page does not contain it, nothing else matters.

Verifying, not assuming

Several of these rules exist because a confident-sounding claim stays false until someone measures it — a schema type that looks real and returns 404, a cache-control header nobody inspected, a sitemap written during a release phase to a container that is then discarded. When a skill here tells you to verify something with curl or a test, that is why.

MIT.